International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-5, Issue-12, Dec-2019]

https://dx.doi.org/10.22161/ijaems.512.2 ISSN: 2454-1311

A voltage sensitivity index application for power

system load shedding considering the generator
controls
L T Nghia1, T T Giang1, Q H Anh1, P T T Binh2, N T An1, P H Hau1
1Department of Electrical and Electronics Engineering, HCMC University of Technology and Education, Vietnam
Email: trongnghia@hcmute.edu.vn
2Department of Electrical and Electronics Engineering, HCMC University of Technology, Vietnam

Email: thanhbinh055@yahoo.com

Abstract—This paper proposes a method for calculating the minimum amount of power load needed to shed and
distributing it for each load bus in order to recover the frequency and voltage back to the allowable range.
Based on the consideration of the primary control of the turbine governor and the reserve power of the
generators for secondary control, the minimum amount of load shedding was calculated in order to recover the
frequency of the power system. Computation and analysis of Voltage Sensitivity Index (VSI) of the load bus to
prioritize distribution of the amount power load shedding at these positions. The lower the load bus have the
Voltage Sensitivity Index (VSI), the higher the amount of load shedding will shed and vice versa. With this
technique, frequency and voltage value are still within allowable range, and a large amount of load shedding
could be avoided, hence, saved from economic losses, and customer service interruption. The effectiveness of the
proposed method tested on the IEEE 37 bus 9 generators power system standard has demonstrated the
effectiveness of this method.
Keywords— Load shedding, Voltage Sensitivity Index (VSI), Frequency control, Primary control, Secondary
control.

I. INTRODUCTION amount load shedding, which based on the frequency

The imbalance active power between the generation and
the load demand causes a decrease the frequency in the
power system. The monitoring and control system will
immediately implement the control solutions to restore the
frequency back to the allowable value, and improve the
power system stability [1]. In [2], the primary and
secondary control power plants are set by automatic
controlled equipment or the power system operator. After
implementing all possible control solutions that the
system's frequency has not yet recovered to the allowable
value, the most efficient method is to reduce the load [3].
The under-frequency load shedding relay (UFLS) is the
traditional load shedding method used quite commonly in
the current power system. In [4], the relays are set to
operate whenever the frequency drops to a specified level
and a fixed amount of load power is shed to restore the
frequency. Using under frequency load shedding relay to
disconnect the load bus will make in insufficient or
excessive load shedding and take a long time to restore the
frequency back to stable. This result will make damages
for the suppliers and customers using the system's power.
The authors in [5], [6] showed methods to estimate the
reduction, or the rate of change of frequency (ROCOF).
The combination of Intelligent load shedding methods has
also been studied and developed such as Artificial Neural
Network (ANN) [7], fuzzy logic algorithms [8], genetic
algorithm (GA) [9] or particle swarm optimization (PSO)
algorithm. These methods minimize the load shedding
costs at steady state operation of pow system [10], [11].
A good load shedding program should be shed with the
minimum number of load buses and as quickly as possible,
and must meet the system's permissible frequency
conditions. On the other hand, in large disturbances of the
power system, frequency decay is often associated with
voltage decay. Voltage decay at load buses reduces system
load, therefore the reduction in frequency is slowed and
the actual load shedding by UFLS is reduced relative to the
level required [12] - [14]. In most practical conditions, the
amount of load shedding may be more or less the amount
of power the system needs to maintain the frequency,
which may lead to cost losses as well as affects the objects
that the system serves.
This paper presents a new method of load shedding to
minimize the amount of load shedding power. The load

www.ijaems.com Page | 618

International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-5, Issue-12, Dec-2019]
https://dx.doi.org/10.22161/ijaems.512.2 ISSN: 2454-1311

shedding strategies based on the Voltage Sensitivity Index

(VSI) to find out the priority and distribute the amount of
load shedding power for each load bus. For load buses, the
lower the load bus have the Voltage Sensitivity Index
(VSI), the higher the amount of load shedding will shed
and vice versa.er system [10], [11].

II. METHODOLOGY
2.1 Overview the power systerm frequency responds
The basic concepts of speed governing are best illustrated
by considering an isolated generating unit supplying a
local load as shown in Figure 1.
Fig. 3: The effects of the governor speed droop and
the frequency of load on the net frequency [18].
The transfer function relating the load
change, PL , to the frequency change,  , is
 1 
Fig. 1: Generator provides independent load  
f ( s )  PL ( s )  Ms  D 
The power system loads are a composite of a variety of  1 1  1  1 
electrical devices. For resistive loads, such as lighting and 1     
heating loads, the electrical power is independent of  R  1  sTG  1  sTCH   Ms  D  

frequency. In the case of motor loads, such as fans and (2)

pumps, the electrical power changes with frequency due to Where: KG the amplification stage; ref reference speed;
changes in motor speed. The references [15], [16] showed
the overall frequency-dependent characteristic of a
TCH “charging time” time constant; PValve per unit change
composite load: in valve position from nominal; Ms angular momentum of
Pe  PL  D the machine in Laplace transform; R is equal to pu change
Nonfrequency  sensitive  load  change Frequency  sensitive  load  change in frequency divided by pu change in unit output; it is
(1)
characteristic for the sliding speed
where: PL - is the load component does not depend on
adjustment; R  f / P
frequency, eg heat load, lighting, ...; D r - the change The purpose of system simulation in the form of a transfer
in load depends on the change of frequency, eg, motors, function is to calculate the time response of the frequency
pumps, etc; Deviation of power change; : Deviation deviation when the load change step is ΔP L. From the
of angle speed change; D: The percentage change in load above description, frequency deviation in steady state it
with percentage of change in frequency varies, D is from 1 means the values of the transfer function is determined for
÷ 2%. s = 0:
The transfer function block diagram reflects the The steady-state value of f ( s ) may be found by:
relationship between the load change and the frequency 1
taking into account the governor characteristic, the prime PL  
f steady state = lim  sf ( s) =  D   PL (3)
mover and the load response shown in Figure 2 [17]. s 0  1  1  1  D
1    
 R  D  R
When the power system has multiple generators with
independent governors, the frequency deviation in steady
state is calculated according to formula (4).
PL (4)
f 
1 1 1
 ...  D
R1 R2 Rn

Fig. 2: The transfer function block diagram describes the

relationship between the load changes and frequency

www.ijaems.com Page | 619

International Journal of Advanced Engineering, Management and Science (IJAEMS)
https://dx.doi.org/10.22161/ijaems.512.2
1
Set    1  D  is the general frequency response
R 
 eq  Compared with the case the generators do not have a
characteristic of power system. It includes the adjustment
characteristics of turbine mechanical power and load.
From formula (4), obtain:
f  PL . (5)
2.2 Primary and secondary frequency control in the
power system
Primary frequency control is an instantaneous adjustment
process performed by a large number of generators with a
turbine power control unit according to the frequency
variation. Secondary frequency control is the subsequent
adjustment of primary frequency control achieved through
the AGC's effect (Automatic Generation Control) on a
number of units specifically designed to restore the
frequency back to its nominal value or otherwise, the
frequency-adjusting effects are independent of the
governor's response called the secondary frequency
control. The process of the primary and secondary
frequency control was shown in Figure 4.
Fig. 4: The relationship between frequency deviation and
output power deviation.
Characteristic line (A) shows the effect of the governors:
change the turbine power according to the change of
frequency:
In balance mode, the intersection of the generator
characteristic line (A) with the frequency characteristic of
the load line (D) determines the frequency f0 equal 50Hz
(or 60 Hz). When the load increases ΔPL, the new
characteristic line will be line (E): P t+∆P. In addition, the
intersection of the generator characteristic line (A) with the
new load characteristic line (E) defines the new frequency
f1. Here, f1< f0. Compared to the case where the generator
does not have a governor, characteristic line (B), it is clear
that: f1< f0. According to the characteristic line (A) of the
generator unit, the governor does not prevent the
frequency reduction: ∆f=f0-f1
www.ijaems.com
[Vol-5, Issue-12, Dec-2019]
ISSN: 2454-1311
However, because the generator has the governor, it has
helped to limit the large deviation of the frequency.
governor (B), the intersection with the new characteristic
line of the load (E) will determine the frequency f1:
f’1<f1<f0. Thus, the governor of the generator unit has the
effect of reducing the large change of frequency known as
the primary frequency controller. The efficiency of the
primary frequency control depends on the slope of the
speed-droop characteristic of the generator units. In the
ideal case, the adjusting characteristic line (F) of the
generator unit is vertical line; the frequency will not
change until the power limit of the generator unit P n.
The above characteristics of the primary adjustment
process lead to the need for external intervention (by the
automatic control device or by the power system
operators) - that is the secondary frequency control
process. The secondary adjustment characteristic is
represented by the simultaneous shifting of the
characteristic line (A) into the characteristic line (C) of the
generator unit, with the slope unchanged. This adjustment
is equivalent to the creation of a static vertical adjustment
characteristic line (F). Thus, the secondary adjustment is
within the rated power range of the generator unit to
restore and maintain the frequency within the allowable
value.
2.3 Calculate the minimum load shedding power
considering the control characteristic of turbine
mechanical power and load
In the 60Hz power system, the frequency deviation
allowed ∆fp is 0.3 Hz (∆fp = -0.3Hz). In the case of the
power deficiency or imbalance between the source and the
load causing the frequency difference, the frequency
control will be implemented in the following order:
primary control, secondary control. When the reserved
power is used for secondary control and the frequency has
not been restored to the permitted value, the load will be
shed. Thus, from formula (5), the relationship between the
permissible change in frequency, the amount of secondary
control power and the minimum load shedding power
PLSmin is calculated according to the proposed formula
below:
f p    .[PL  (PSecondary control  PLS min )]
(6)
In this case, if (∆PSecondary control + ∆PLS min) < ∆PSacondary max,
then ∆PLSmin=0, otherwise the minimum power load
shedding is calculated by the formula below:
 f p 
PLS min  PL     PSecondary Max
   (7)
Where: ∆fp is the permissible change in frequency (pu);
PLSmin is the minimum amount of power required to shed
Page | 620
 International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-5, Issue-12, Dec-2019]
https://dx.doi.org/10.22161/ijaems.512.2 ISSN: 2454-1311

(pu); ∆PSecondary control is the amount of secondary control The effectiveness of the proposed method is tested on the
power addition to the system. IEEE 37 bus 9 generators system [21] which is shown in
2.4 The Voltage Sensitivity Index (VSI) Figure 5. Total the active power and the reactive power of
The main objective of calculating VSI is to find the most the system are 1024.31 MW and 215.94 MVAR
sensitive node of the system from voltage sensitivity point respectively under normal operating conditions. The
of view [19]. It considered a numerical solution, which maximum active power and reactive power of the system
helps operator to monitor how to shed the load to prevent are 1087 MW and 449 MVAR. The control solutions
frequency and voltage collapse. Nodes, having minimum minimize the amount of load shedding and maintain
voltage sensitivity index are selected and then, using steady-state frequency from 59.7 to 60 Hz. To test the
equation (8) to calculate the voltage sensitivity index effectiveness of the proposed method, the outage situations
(VSI). of the generator units are calculated, simulated and tested
VSI at bus i, is defined as [20]: the parameters. In the case of calculations and simulations,
n the spinning reserved power to control the secondary
 (1  V ) k
2
frequency is also considered. All test cases are simulated
VSI i  k 1 on PowerWorld GSO 19 software. The results are
n (8) compared with the results of the traditional load shedding
th
Where Vk is voltage at k node and n is the number of method using under frequency load shedding relay.
nodes. Apply the (5), (6), (7) formulas calculate the system
After calculating the minimum amount of load shedding frequency, the amount of primary and secondary control
power required, the next issue determines the distribution power and the amount of load to be shed. The results of
the amount of load shedding power at load buses. The the computation of the outage generator situations are
node with least VSI will be picked as the best location for shown in Table 1.
the load shedding. That mean, the amount of load shedding
at different load buses can be calculated in the same way
as the principle of the load sharing in the parallel circuit.
Due to parallel electrical circuits:
1 1 1 1
Y    ... 
Z eq Z1 Z 2 Zn
(9)
1
Z eq 
Y (10)
2
P1 U Z eq Z eq
 . 
P Z1 U 2 Z1 (11)
Equivalent formula follows: Fig. 5: The IEEE 37 bus 9 generators test system.
Pi Z eq VSI eq

P Zi VSI i Table 1. The outage generators cases
(12) Name of Freque In the The The The
Here, the general formula calculates the amount of load Gen. Bus ncy allow primary second amount
shedding distribution at nodes according to the Voltage (Hz) range control ary of load
VSI eq power control sheddi
Sensitivity Index (VSI): PLSi  .PLS min (13)
VSI i value power ng
(MW) value (MW)
Where, n is the number of generator bus; i is the number of (MW)
load bus; PLSi the amount of load shedding power for the i REDBUD 59.97 Yes 10 0 0
bus (MW); PLS min the minimum amount of load shedding ELM345# 59.56 No 150 12.5 27.77
power to the restore of frequency back to the allowable 1
value (MW); VSIi: the Voltage Sensitivity Index (VSI) of ELM345# 59.56 No 150 12.5 27.77
the i bus; VSIeq: the equivalent Voltage Sensitivity Index 2
(VSI) of all load buses. PEACH69 59.62 No 112.31 16.32 11.89
III. CASE STUDIES - SIMULATION AND RESULTS CEDAR69 59.86 Yes 52 0 0
BIRCH69 59.79 Yes 80 0 0

www.ijaems.com Page | 621

International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-5, Issue-12, Dec-2019]
https://dx.doi.org/10.22161/ijaems.512.2 ISSN: 2454-1311

PEAR138 59.6 No 140 17.38 15.91 MW. The frequency of the system after the
PEAR69 59.7 Yes 110 0 0 implementation of the secondary control is shown in
In the test example, the sudden disconnection of the Figure 6.
PEAR138#1 generator (bus 53) is simulated. Applying the Thus, after performing the secondary control, the recovery
equation (5) calculates the stable frequency value when the frequency is 59.65 Hz and has not yet returned to the
PEAR138#1 generator (bus 53) disconnects from the allowed value. Therefore, the final solution is load
system. The frequency value is 59.57 Hz, and shows in shedding. Equation (7) is applied to calculate the minimum
Figure 6. amount of load shedding power to recovery the frequency
in allowable value.
 f p 
PLS min  PL     PSecondary Max
  
 0.3 
 1.4     0.1738  0.1591 pu  15.91MW
 0.00482 x 60 

After calculating the minimum load shedding power, the

load shedding distribution at the load buses is calculated.
The amount of load shedding at load buses based on the
Voltage Sensitivity Index (VSI) value of all load buses.
Calculation steps at section II.4 are applied to calculate the
Voltage Sensitivity Index (VSI) value of all buses. The
Fig. 6: The frequency of the system when the PEAR138#1
Voltage Sensitivity Index (VSI) value and the voltage at
generator disconnects
load bus (pu) at all load buses is shown in Figure 7.
After the PEAR138#1 generator suddenly disconnects, the
The priority load shedding distribution for each load bus is
frequency value is less than the allowable value.
calculated based on the following principle: The lower
Therefore, the primary control and secondary frequency
VSI, the greater the amount of shedding power. Equation
control which presented in section II.2 for frequency
(7) in Section II.3 is applied to calculate the amount of
recovery should be implemented.
disconnection power value at the load buses.
The primary control process is done automatically by the
turbine governor after the PEAR138#1 outage generator.
The value of the primary control power of each generator
turbine is shown in Table 2.
Table 2. The value of the primary control power of the
generators
The increased primary
Generator control power of each
generator (MW)
REDBUD69 (bus 14) 5.2
ELM345#1 (bus 28) 23.6
ELM345#2 (bus 28) 23.6
SLACK345 (bus 31) 32.8
PEACH69 (bus 44) 21
CEDAR69 (bus 48) 7.5
BIRCH69 (bus 50) 11.2
PEAR138 (bus 53) 0
PEAR69 (bus 54) 15.1
Total = 140 MW Fig.7. The VSI and the voltage at load buses (pu) after the
Because the recovery frequency is less than the allowable outage generator (PEAR138)
value, so the secondary control is implemented after the In order to compare the effectiveness of the proposed
primary control. The spinning reversed power of the method, the load shedding method using under frequency
generators will be mobilized to perform the secondary load shedding relay is used to compare. The process of
control. In this case, the secondary control power is 17.38 UFLS is implemented when the frequency reduces below

www.ijaems.com Page | 622

International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-5, Issue-12, Dec-2019]
https://dx.doi.org/10.22161/ijaems.512.2 ISSN: 2454-1311

the frequency setting threshold. The load is usually cut

step-by-step based on the load shedding table that pre-
designed based on the general rule and operator
experience. These tables guide the amount of load that
should be cut at each step depending on the decrease of
frequency. These values are shown in Table 3.
Table 3. The UFLS scheme using load shedding table[22]
The amount
Total
of load Fig.9: The voltage at all bus when load shedding
The Time amount
Frequency shedding (the according to proposed method
steps delay of load
(Hz) percent of
UFLS (s) shedding
total load)
(%)
(%)
A 59.7 0.28 9 9
B 59.4 0.28 7 16
C 59.1 0.28 7 23
D 58.8 0.28 6 29
E 58.5 0.28 5 34
F 58.2 0.28 7 41
J 59.4 10 5 46 Fig.10: The voltage at all bus when load shedding
The frequency and the rotor angle comparison between the according to the UFLS method
proposed method and the UFLS method are presented in
Figure 8. IV. CONCLUSION
It can be seen that the proposed load shedding method has A load shedding method considers to the primary and
less the amount of shedding (76.28 MW) than the UFLS. secondary control elements of the power plant to calculate
Here, the recovery frequency value of the proposed the minimum amount of load shedding power and restore
method is lower than the UFLS method. However, this the frequency back to the allowable value. The proposed
value is still within allowable parameter and acceptable method ensures the frequency and voltage of the system in
range (59.7Hz). Especially, when considering the voltage case of a severe generation–load mismatch. The selection
value of the proposed method is equivalent to the UFLS of location and distribution of load shedding power at load
method, although this method has less the amount of load buses are based on the Voltage Sensitivity Index (VSI)
shedding than UFLS method. This can be explained by the concept. The effectiveness of the proposed method has
fact that a large load at load nodes with the lower VSI are been demonstrated on a 9-machine, 37-bus system under
disconnected causing the voltage to recover faster. Figure test cases. The performance of this method is found to be
9 and Figure 10 show that the voltage at all nodes after better than that of a conventional UFLS scheme. The test
load shedding according to the proposed method and the results show that the proposed method results in reduced
UFLS method is near the same. amount of load shedding while satisfying the operating
conditions and limitations of the network.

ACKNOWLEDGEMENTS
This research was supported by the HCMC University of
Technology and Education.

Fig. 8: The frequency comparison between the proposed
method and the traditional method
REFERENCES
[1] Raja Masood Larik, Mohd.Wazir Mustafa and Manoj
Kumar Panjwani, “A statistical jacobian application for
power system optimization of voltage stability,” Indonesian
Journal of Electrical Engineering and Computer Science,
vol. 13, No. 1, 2019, pp. 331–338.
[2] Sam Weckx, Reinhilde D'Hulst and Johan Driesen,
“Primary and Secondary Frequency Support by a Multi-
Agent Demand Control System,” IEEE Transactions on
Power Systems, vol. 30, pp. 1394-1404, May 2015.

www.ijaems.com Page | 623

International Journal of Advanced Engineering, Management and Science (IJAEMS) [Vol-5, Issue-12, Dec-2019]
https://dx.doi.org/10.22161/ijaems.512.2 ISSN: 2454-1311

[3] Raghu C. N. and A. Manjunatha, “Assessing Effectiveness International Journal of Applied Power Engineering
of Research for Load Shedding in Power System,” (IJAPE), Vol.7, No.1, pps 18-26, April 2018.
International Journal of Electrical and Computer [17] Allen J. Wood, Bruce F. Wollenberg and Gerald B. Sheblé,
Engineering (IJECE), vol. 7, pp. 3235–3245, Dec 2017. “Power Generation, Operation and Control,” Third Edition,
[4] Gazmend Kabashi and Skender Kabashi, “Review of under John Wiley & Sons, Inc, pp. 473-481, 2014.
Frequency Load Shedding Program of Kosovo Power [18] A. Goikoetxea, J.A. Barrena, M.A. Rodriguez and F.J.
System based on ENTSO-E Requirements,” International Chivite, “Frequency restoration in insular grids using
Journal of Electrical and Computer Engineering (IJECE), Ultracaps ESS,” Power Electronics Electrical Drives
vol. 8, pp. 741–748, April 2018. Automation and Motion (SPEEDAM) International
[5] Lukas Sigrist, “A UFLS Scheme for Small Isolated Power Symposium on, pp. 1280-1285, July 2010.
Systems Using Rate-of-Change of Frequency”, IEEE [19] Qian Kejun, Zhou Chengke, allan Malcolm, Y uan Yue.
Transactions on Power Systems, Vol: 30, Issue: 4, pp. 2192 “Effect of load models onassessment of energy losses in
– 2193, 2015. distribution generation planning,”. Electr Power Res,
[6] Turaj Amraee, Mohammad Ghaderi Darebaghi, Alireza 2:1243–50, 2011.
Soroudi, Andrew Keane, “Probabilistic Under Frequency [20] Gopiya Naik S., D. K. Khatod and M. P. Sharma. “Optimal
Load Shedding Considering RoCoF Relays of Distributed Allocation of Distributed Generation in Distribution System
Generators”, IEEE Transactions on Power Systems, Vol: 33, for Loss Reduction,” IPCSIT vol. 28, 2012.
pp. 3587 – 3598, 2018. [21] J. Duncan Glover, Mulukutla S. Sarma and Thomas J.
[7] C. T. Hsu, M. S. Kang and C. S. Chen, “Design of Adaptive Overbye., “Power System Analysis and Design,” Fifth
Load Shedding by Artificial Neural Networks”, IEE Edition, Cengage Learning, 2012. The 37 bus-9 generators
Generation, Transmission, Distribution, Vol. 152, Issude: 3, test system (Ex.11.9). Available online:
pp. 415-421, May 2005. https://www.powerworld.com/simulator-19-glover-sarma-
[8] A. A. Sallam, and A. M. Khafaga, “Fuzzy Expert System overbye-edition-download(accessed on 15 January 2019)
Using Load Shedding for Voltage Instability Control,” IEEE [22] Florida Reliability Coordinating Council, “FRCC Regional
Large Engineering Con. on Power Engineering, June 2002. Underfrequency Load Shedding (UFLS) Implementation
[9] W. P. Luan, M. R. Irving, and J. S. Daniel, “Genetic Schedule, FRCC handbook,” June 27, 2011.
Algorithm for Supply Restoration and Optimal Load
Shedding in Power System Distribution Networks”, IEE
Proceedings - Generation, Transmission and Distribution,
Vol. 149, Issue: 2, pp. 145-51, March 2002.
[10] Kantilal Dayalal Joshi, Vinod Chandrakar, “Transient
Stability Improvement Using UPFC-SMES in A Multi
Machine Power System,” International Journal of Applied
Power Engineering (IJAPE), Vol.5, No.1, pp. 14-21, April
2016.
[11] G. A. M. Hosaini Hajivar, S. S. Mortazavi, “Impact of Shunt
FACTS Devices on Security Constrained Unit
Commitment,”, International Journal of Applied Power
Engineering (IJAPE), Vol. 5, No. 1, pp. 22-39, April 2016.
[12] P. Pinceti, M. Vanti, C. Brocca, M. Carnesecchi and G.
Macera, “Design criteria for a power management system
for microgrids with renewable sources,” Electr. Power Syst.
Res, vol 122, pp. 168–179, May 2015.
[13] K. Balasubramaniam, P. Saraf, R. Hadidi and E.B. Makram,
“Energy management system for enhanced resiliency of
microgrids during islanded operation,” Electr. Power Syst.
Res vol 137, pp. 133–141, August 2016.
[14] M. Moafi, M. Marzband, M. Savaghebi and J.M. Guerrero,
“Energy management system based on fuzzy fractional
order PID controller for transient stability improvement in
microgrids with energy storage,” Int. Trans. Electr. Energy
Syst, vol 26 (3), pp. 1-20, Feb 2016.
[15] PrabhaKundur, “Power System Stability and Control,” First
Edition, McGraw-Hill Inc, pp. 507, 1994.
[16] C. R. Balamurugan, “Three Area Power System Load
Frequency Control Using Fuzzy Logic Controller”,

www.ijaems.com Page | 624